Using sentiment analysis to forecast stock short-term trend

Tan, Lin (2024) Using sentiment analysis to forecast stock short-term trend. Final Year Project, UTAR.

Preview

PDF Download (4Mb) | Preview

Abstract

This research investigates the effectiveness of sentiment analysis in stock market prediction, integrating advanced computational techniques with financial analytics. Specifically, the study examines the efficacy of the Autoregressive Distributed Lag (ARDL) model combined with the GPT-4 Turbo model from OpenAI for sentiment analysis to predict the stock price movements influenced by various news sources in Malaysia. The project employs a systematic methodology to preprocess data, integrate sentiment scores, and apply the ARDL model to analyze the impact of news sentiment on stock prices. The sentiment analysis, powered by GPT-4 Turbo, provides a robust framework for interpreting the emotional tone within financial news content. Results indicate that the ARDL model, while capturing general market trends and oscillations, exhibits moderate success in forecasting, as evidenced by varying RMSE values across different news sources. This variability highlights the influential capacity of news sources and underscores the necessity for nuanced analysis techniques. The findings contribute to the broader understanding of how different news sources impact stock market movements and demonstrate the potential for enhanced predictive accuracy through the integration of advanced AI-driven tools in financial forecasting. The study’s insights encourage further exploration into hybrid models that might combine traditional financial indicators with innovative sentiment analysis methodologies to improve the reliability and effectiveness of stock market predictions.

Actions (login required)